Spaces:
Running
on
CPU Upgrade
Running
on
CPU Upgrade
fix: fix the bug in unselecting anonymous
Browse files- src/utils.py +3 -3
- tests/test_utils.py +22 -2
src/utils.py
CHANGED
@@ -113,10 +113,10 @@ def select_columns(
|
|
113 |
selected_cols.append(c)
|
114 |
# We use COLS to maintain sorting
|
115 |
filtered_df = df[FIXED_COLS + selected_cols]
|
116 |
-
filtered_df[COL_NAME_AVG] = filtered_df[selected_cols].apply(calculate_mean, axis=1).round(decimals=2)
|
117 |
-
filtered_df.sort_values(by=[COL_NAME_AVG], ascending=False, inplace=True)
|
118 |
-
filtered_df.reset_index(inplace=True, drop=True)
|
119 |
if reset_ranking:
|
|
|
|
|
|
|
120 |
filtered_df[COL_NAME_RANK] = filtered_df[COL_NAME_AVG].rank(ascending=False, method="min")
|
121 |
|
122 |
return filtered_df
|
|
|
113 |
selected_cols.append(c)
|
114 |
# We use COLS to maintain sorting
|
115 |
filtered_df = df[FIXED_COLS + selected_cols]
|
|
|
|
|
|
|
116 |
if reset_ranking:
|
117 |
+
filtered_df[COL_NAME_AVG] = filtered_df[selected_cols].apply(calculate_mean, axis=1).round(decimals=2)
|
118 |
+
filtered_df.sort_values(by=[COL_NAME_AVG], ascending=False, inplace=True)
|
119 |
+
filtered_df.reset_index(inplace=True, drop=True)
|
120 |
filtered_df[COL_NAME_RANK] = filtered_df[COL_NAME_AVG].rank(ascending=False, method="min")
|
121 |
|
122 |
return filtered_df
|
tests/test_utils.py
CHANGED
@@ -1,7 +1,8 @@
|
|
1 |
import pandas as pd
|
2 |
import pytest
|
3 |
|
4 |
-
from src.utils import filter_models, search_table, filter_queries, select_columns, update_table_long_doc, get_iso_format_timestamp, get_default_cols
|
|
|
5 |
|
6 |
|
7 |
@pytest.fixture
|
@@ -92,4 +93,23 @@ def test_get_default_cols():
|
|
92 |
cols, types = get_default_cols("qa")
|
93 |
for c, t in zip(cols, types):
|
94 |
print(f"type({c}): {t}")
|
95 |
-
assert len(frozenset(cols)) == len(cols)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import pandas as pd
|
2 |
import pytest
|
3 |
|
4 |
+
from src.utils import filter_models, search_table, filter_queries, select_columns, update_table_long_doc, get_iso_format_timestamp, get_default_cols, update_table
|
5 |
+
from src.display.utils import COL_NAME_IS_ANONYMOUS, COL_NAME_REVISION, COL_NAME_TIMESTAMP, COL_NAME_RERANKING_MODEL, COL_NAME_RETRIEVAL_MODEL, COL_NAME_RANK, COL_NAME_AVG
|
6 |
|
7 |
|
8 |
@pytest.fixture
|
|
|
93 |
cols, types = get_default_cols("qa")
|
94 |
for c, t in zip(cols, types):
|
95 |
print(f"type({c}): {t}")
|
96 |
+
assert len(frozenset(cols)) == len(cols)
|
97 |
+
|
98 |
+
|
99 |
+
def test_update_table():
|
100 |
+
df = pd.DataFrame(
|
101 |
+
{
|
102 |
+
COL_NAME_IS_ANONYMOUS: [False, False, False],
|
103 |
+
COL_NAME_REVISION: ["a1", "a2", "a3"],
|
104 |
+
COL_NAME_TIMESTAMP: ["2024-05-12T12:24:02Z"] * 3,
|
105 |
+
COL_NAME_RERANKING_MODEL: ["NoReranker"] * 3,
|
106 |
+
COL_NAME_RETRIEVAL_MODEL: ["Foo"] * 3,
|
107 |
+
COL_NAME_RANK: [1, 2, 3],
|
108 |
+
COL_NAME_AVG: [0.1, 0.2, 0.3], # unsorted values
|
109 |
+
"wiki_en": [0.1, 0.2, 0.3]
|
110 |
+
}
|
111 |
+
)
|
112 |
+
results = update_table(df, "wiki", "en", ["NoReranker"], "", show_anonymous=False, reset_ranking=False, show_revision_and_timestamp=False)
|
113 |
+
# keep the RANK as the same regardless of the unsorted averages
|
114 |
+
assert results[COL_NAME_RANK].to_list() == [1, 2, 3]
|
115 |
+
|